I completely agree. But then again, it's a very clever way of achieving matrix multiplication in a map-reduce fashion.
Apologies for the brevity, this was sent from my iPhone On Jan 24, 2011, at 10:33, Ted Dunning <ted.dunn...@gmail.com> wrote: > Yes. It is the intended behavior, but it is damned confusing to new comers! > > On Mon, Jan 24, 2011 at 3:37 AM, Shannon Quinn <squ...@gatech.edu> wrote: > >> I think this is actually the intended behavior? See the javadoc comments >> for times(). >> >> Apologies for the brevity, this was sent from my iPhone >> >> On Jan 24, 2011, at 6:27, "Ahmed Nagy (JIRA)" <j...@apache.org> wrote: >> >>> Multiplication of two matrices that are distributed one of them is >> transpose A*A' gives the result of A*A >>> >> ---------------------------------------------------------------------------------------------------------- >>> >>> Key: MAHOUT-592 >>> URL: https://issues.apache.org/jira/browse/MAHOUT-592 >>> Project: Mahout >>> Issue Type: Bug >>> Components: Math >>> Affects Versions: 0.5 >>> Environment: Operating System: Ubuntu 10.10 >>> Application Mahout 0.5 trunk >>> Program used to run the Application I a developing is eclipse Version: >> 3.5.2 >>> Hadoop 0. 20.2 >>> Reporter: Ahmed Nagy >>> Priority: Blocker >>> >>> >>> The problem basically is the result of two distributed matrices A and A' >> gives always A * A. I am attaching the code that produces the issue. If A is >>> 1.000 2.000 3.000 >>> 4.000 5.000 6.000 >>> 7.000 8.000 9.000 >>> B=A' >>> 1.000 4.000 7.000 >>> 2.000 5.000 8.000 >>> 3.000 6.000 9.000 >>> Result should be >>> 14.000 32.000 50.000 >>> 32.000 77.000 122.000 >>> 50.000 122.000 194.000 >>> But Result is >>> 30.0 36.0 42.0 >>> 66.0 81.0 96.0 >>> 102.0 126.0 150.0 >>> which is the result of A*A >>> >>> -- >>> This message is automatically generated by JIRA. >>> - >>> You can reply to this email to add a comment to the issue online. >>> >>
